This thesis addresses the challenge of managing complex engineering knowledge in Advanced Driver Assistance Systems (ADAS) by introducing a multi-agentic system for automated Knowledge Graph (KG) construction from unstructured technical texts. This system employs an agent-inspired pipeline where specialized LLM-powered agents for information extraction and normalization collaborate with deterministic modules for validation, ensuring the semantic consistency and structural integrity of the resulting KG. A core contribution is the integration of an embedding-based "Commonsense Verifier" to assess the plausibility of extracted facts and an "Inductive Reasoner" to enrich the KG with inferred knowledge. Evaluation demonstrated the system's end-to-end functionality, showing it effectively filters implausible information and manages redundancy, thereby validating that an orchestrated system of LLM agents and deterministic checks can create a more robust and coherent knowledge base for safety-critical domains.